Backtracking orthogonal least squares algorithm for model selection
Backtracking orthogonal least squares algorithm for model selection
The orthogonal least squares (OLS) algorithm [l] is an efficient implementation of the forward-selection [2] method for subset model selection. The ability to find good subset parameters with only a linearly increasing computational requirement makes this method attractive for practical implementations. This technique, however, has been criticised [2] as it does not guarantee to find an optimum solution. In this paper, we examine why forward-selection technique can fail to find optimum subset models and present a modification scheme to improve the selection process.
10/1-10/5
Chng, E. S.
fea46228-47bf-4963-83d7-b7b5591183de
Mulgrew, B.
95a3fbda-7de2-4583-b1f2-0a54a69b414a
Chen, S.
9310a111-f79a-48b8-98c7-383ca93cbb80
10 February 1994
Chng, E. S.
fea46228-47bf-4963-83d7-b7b5591183de
Mulgrew, B.
95a3fbda-7de2-4583-b1f2-0a54a69b414a
Chen, S.
9310a111-f79a-48b8-98c7-383ca93cbb80
Chng, E. S., Mulgrew, B. and Chen, S.
(1994)
Backtracking orthogonal least squares algorithm for model selection.
IEE Colloquium on Maths. Aspects of Digital Signal Processing, Bristol, United Kingdom.
.
Record type:
Conference or Workshop Item
(Paper)
Abstract
The orthogonal least squares (OLS) algorithm [l] is an efficient implementation of the forward-selection [2] method for subset model selection. The ability to find good subset parameters with only a linearly increasing computational requirement makes this method attractive for practical implementations. This technique, however, has been criticised [2] as it does not guarantee to find an optimum solution. In this paper, we examine why forward-selection technique can fail to find optimum subset models and present a modification scheme to improve the selection process.
Text
c-cmdsp1994
- Author's Original
Restricted to Repository staff only
Request a copy
More information
Published date: 10 February 1994
Additional Information:
IEE Colloquium on Maths. Aspects of Digital Signal Processing (Bristol, UK), Feb. 10, 1994. Event Dates: Feb. 10, 1994 Organisation: IEE
Venue - Dates:
IEE Colloquium on Maths. Aspects of Digital Signal Processing, Bristol, United Kingdom, 1994-02-10
Organisations:
Southampton Wireless Group
Identifiers
Local EPrints ID: 251094
URI: http://eprints.soton.ac.uk/id/eprint/251094
PURE UUID: 1850b9ec-4e41-41a0-8cb3-64bea691e0c8
Catalogue record
Date deposited: 12 Oct 1999
Last modified: 14 Mar 2024 05:09
Export record
Contributors
Author:
E. S. Chng
Author:
B. Mulgrew
Author:
S. Chen
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics